Prehospital lung ultrasound in acute heart failure: Impact on diagnosis and treatment

Patients with acute heart failure (AHF) are commonly misdiagnosed and undertreated in the prehospital setting. These delays in diagnosis and treatment have a direct negative impact on patient outcomes. The goal of this study was to determine the diagnostic accuracy of paramedics with and without the use of lung ultrasound (LUS) for the diagnosis of AHF in patients with dyspnea in the prehospital setting. Secondarily, we assessed LUS impact on rate of and time to initiation of HF therapies.


INTRODUC TI ON
Acute heart failure (AHF) is a clinical syndrome characterized by new onset of heart failure or worsening heart failure presenting with cardiorespiratory symptoms. 1In the United States, nearly 1 million hospitalizations occur each year from AHF. 2 The 1-year hospitalization and 1-year mortality rates of heart failure were 25.9% and 26.7%, respectively. 1These poor outcomes are thought to be a result of delays in diagnosis and treatment initation. 3Early treatment, including in the prehospital setting, is directly correlated with decrease length of hospital stay and mortality. 3,4F has classically been diagnosed in the emergency department (ED) using patient history, physical examination, chest radiography, B-type natriuretic peptide, and Doppler echocardiography. 5,6wever, studies have shown that lung ultrasound (LUS) is more accurate for diagnosing AHF. 7,8LUS, through assessment of B-lines, is an accurate and easy-to-use tool used for the detection of pulmonary congestion seen in AHF patients.Ultrasound has been adapted in the prehospital setting for decades in other countries. 9In the United States, LUS has been studied in the prehospital setting, [10][11][12][13][14][15][16] with varying results on the diagnostic accuracy for AHF and to our knowledge only one prior study exploring the impact of diagnosis on early management. 13e goal of this initial pilot study was to determine the paramedic diagnostic accuracy for detecting AHF in short of breath patients transported by emergency medical services (EMS) with and without the use of LUS.Secondarily, we assessed how paramedic performed LUS impacted rate and timing of HF therapy.

Study setting and population
This was a prospective, nonrandomized interventional study on a consecutive sample of patients transported to the hospital by one EMS agency that responds to around 14,000 calls annually.The study was conducted from March to August 2022.This study was approved by the local institutional review board.

Study protocol
Adult patients (>18 years old) with an on-scene chief complaint of dyspnea and at least one of the following-bilateral lower extremity edema, orthopnea, wheezing or rales on auscultation, respiratory rate > 20 breaths/min, or oxygen saturation < 92%--were included.
These criteria were adapted from prior studies. 11,12LUS was performed when a paramedic trained in LUS was present and an ultrasound was available on the ambulance.Ambulances were assigned to 9-1-1 calls based on geographic location and availability regardless of the presence or absence of an ultrasound device.Patients with hypotension (systolic blood pressure < 90 mm Hg), ECG showing ST-elevation, fever >100.4°F,trauma, or pregnancy were excluded.
If patients were transported to a hospital where the electronic medical record could not be accessed, then they were excluded from analysis as well.Exclusion was done in the post hoc data analysis.Twenty-six full-time paramedics, from a single EMS agency with four ambulances, were trained in LUS with 30 min of didactics, 30 min of hands-on scanning, and individual completion of a written knowledge assessment and objective structured clinical examination.Prior to enrollment, each paramedic was required to pass the above assessments and complete an EMS transport with an ultrasound trained faculty or fellow present to demonstrate clinical integration of LUS.
Paramedics with ultrasound training and ultrasound access performed LUS on patients with shortness of breath.If the patient required continuous positive airway pressure (CPAP) this was done prior to the ultrasound to stabilize the patient.LUS was only available on two of the four ambulances during the enrollment period.One ambulance had ultrasound available beginning in March (6 months) the second one beginning in June (3 months).LUS was performed with a Butterfly IQ (Butterfly Network Inc.) handheld ultrasound machine, using the lung preset with a depth of 15 cm (maximum depth the machine allows).Tissue harmonics and multibeam former were turned off.Gain was set by the paramedic for each individual patient to ensure that the pleural line was distinct and the rib shadows were black.Six-second clips were captured.These machine settings were selected based on their use in previous studies. 17,18tients were imaged in a position of comfort either semiupright or in a sitting position, with the transducer in a vertical orientation and the probe indicator toward the patient's head.AHF on LUS was defined as greater than or equal to three B-lines in a zone bilaterally.
Paramedics began by imaging right zone 1 (R1) and left zone 1 (L1) and determined the presence or absence of B-lines.If both R1 and L1 were positive for greater than or equal to three B-lines, then the patient was diagnosed with AHF.If either zone was negative, then zone 4 on that corresponding side was imaged.This imaging protocol was based on previously published simplified LUS protocols for imaging patients with AHF. 19,20deos acquired on patients were stored locally on the ultrasound device and wirelessly uploaded for offline review and quality assurance by an expert in LUS.Paramedics documented LUS findings and their interpretation in real time directly into ESO, the ambulance electronic medical record system.A diagnosis of AHF by paramedics was identified as either the primary or the secondary impression being AHF or pulmonary edema in ESO.
After LUS interpretation, paramedics followed the pulmonary edema treatment pathway if AHF was diagnosed.This treatment pathway included sublingual nitroglycerin and CPAP.Furosemide was not available in the prehospital setting.If AHF was not diagnosed, then paramedics followed their standard treatment pathway based on what they thought was the most likely diagnosis.HF therapy was defined as nitroglycerin administered in the prehospital setting and furosemide or nitroglycerin administered in the ED.CPAP was not included as a predefined HF therapy for the study as it is not a specific treatment for AHF.Time to treatment was defined as paramedic initial patient contact time to the time HF treatment was initiated including in the prehospital setting or the ED.Patients

Data analysis
Sensitivity, specificity, positive likelihood ratios, and negative likelihood ratios with 95% confidence intervals (CIs) were calculated using paramedic diagnosis with and without the use of LUS compared to hospital discharge diagnosis.These were derived using Microsoft Excel and VassarStats (http://vassa rstats.net).Kappa was used to determine LUS interpretation inter-rater reliability between paramedic and expert overread as well as expert to expert.As this was a pilot study no sample size was calculated.
Overall, sensitivity increased by 48% without impacting specificity.Accuracy increased by 13% and the positive likelihood ratio increased by 7.7.There were no significant differences in age, gender, or vital signs between patients who received a LUS versus those who did not (see Table 3).Patients with a final diagnosis of AHF who did not receive LUS had significantly higher past medical histories of hypertension and tobacco use.This was the same when comparing all included patients with and without a LUS performed.LUS median image quality was 4 (range 2 to 5).Inter-rater reliability between paramedic and expert interpretation for the presence of B-lines was k = 0.80 (95% CI 0.61-0.98).Expert-to-expert inter-rater reliability was k = 0.88 (95% CI 0.66-1).

DISCUSS ION
Early diagnosis and accurate treatment are key cornerstones in the management of AHF. 3,4Delays in either diagnosis or treatment are associated with poor outcomes. 3LUS is an accurate tool for identifying AHF through the identification of bilateral B-lines.
In this study we set out to determine the diagnostic accuracy of paramedic-performed LUS for AHF and how this compared to prehospital diagnosis when LUS was not used.With the use of LUS, we found prehospital sensitivity to be moderate and specificity to be high.The use of LUS significantly improved prehospital sensitivity by about 50% without lowering specificity.Additionally, the positive likelihood ratio was 16, suggesting that, in the appropriate clinical context, the presence of bilateral B-lines likely represents AHF and should be treated as such.
Impressively, paramedics with little to no prior training in LUS were able to use this tool accurately in the prehospital setting.This is supported by a high kappa between paramedic and expert interpretation as well as high diagnostic accuracy.Importantly, a modified scanning protocol was utilized to improve compliance and to decrease the amount of time to complete the LUS.This protocol was Despite having a high specificity, our sensitivity was lower than previous studies finding high sensitivity. 10,13There are several reasons our sensitivity was lower than previous studies, and this was an expected finding.We intentionally utilized a limited imaging protocol to increase the likelihood of utilization by paramedics and for simplicity.Because paramedics were only imaging a maximum of four zones, it is possible the nonimaged zones would have been positive for B-lines, thus increasing the sensitivity.Our intention with this study was to have paramedics integrate LUS into their prehospital management of patients with AHF and the investigators felt that by having an eight-zone protocol could potentially delay care or deter providers from using this tool.We anticipated we would not be able to identify every patient with AHF.When comparing to Laursen et al. 10 and Zanatta et al., 13 they both found sensitivities in the mid-90s but had much smaller sample sizes, and LUS procedures were performed by physicians and not paramedics.Both, however, did use a similar scanning protocol.Interestingly, Laursen et al. 10 only found a specificity of 77%.Sensitivity in our study was similar to Schoeneck et al. 12 ; however, their specificity was also low at 72%.This may have been in part to a broad training of 63 paramedics with only one-third using the tool in the field as well as the inability to track images in real time and give feedback to improve usage.In our study, training a smaller cohort of paramedics and the ability to review images almost immediately allowed for prompt feedback and more oversight.Gundersen et al. 16 evaluated 214 patients in the prehospital setting with ultrasound and found similar sensitivity (62%) and specificity (92%) to our study.However, their protocol differed in that they used a combined heart and LUS imaging protocol, and all studies were performed by highly trained physicians comparing preultrasound diagnosis to postultrasound diagnosis.In addition, different ultrasound systems were used in different studies; however, the impact of this difference is unknown.
We found inter-rater reliability between paramedic and expert interpretation to be high and image quality to also be high.This is similar to prior studies evaluating the ability of novices to interpret LUS. 17,22Multiple prior studies have evaluated agreement of paramedic interpretation to expert overread, 11,14,15 with Pietersen et al. also finding high agreement and moderate image quality after 100 paramedics performed 590 ultrasound procedures.However, other studies did not find these promising results with kappa values ranging from 0.17 to 0.44 potentially being related to low uptake by paramedics and poor image quality. 11,15 this study we found that the use of LUS significantly improved the number of patients receiving HF therapy prior to hospital arrival (p = 0.002) and decreasing the time to initiation of HF therapy by over 2 h in the LUS group.This potentially has significant impact on patient outcomes as earlier initiation of HF therapies is associated with lower mortality and hospital readmission rates. 4,23Our study differs from prior studies evaluating LUS in the prehospital setting as the majority have focused on diagnosis and have not evaluated impact on treatment.Zanatta et al. 13 evaluated management changes comparing one ambulance with LUS and one without and found no significant differences in treatment between the two.Their data most likely differs from ours as they used a smaller sample size (60 patients) and their LUS and comparison groups were in completely different practice environments; thus there are many other factors that could have impacted prehospital management.TA B L E 3 AHF patient demographics with and without LUS.

LUS
In our study we did evaluate outcome data including mortality, hospital length of stay, and intubation but did not find any significant difference between the LUS and no ultrasound groups (see Table 4).
This was anticipated as the sample size was too small to evaluate for differences in outcomes.It is important to recognize that despite the fact that a larger percentage of AHF patients were treated with nitroglycerin in the LUS group these patients also received a similar amount of albuterol (p = 0.77) when compared to the group without LUS, which could potentially increase hospital lengths of stay and in-hospital intubations. 24Albuterol administration is challenging to control when administered prior to paramedic arrival.This will be a focus in future studies to decrease unnecessary medication administration.

LI M ITATI O N S
There are several limitations that may limit the generalizability of this study.This was a pilot study conducted at a single EMS agency.
This was by design to determine whether a small group of paramedics could adapt this technology into their prehospital management of patients with dyspnea.Future studies should aim to enroll a larger number of paramedics and from multiple agencies.Paramedics were not blinded to clinical history and examination.This did not likely impact our results, as prior literature has suggested that even when clinical data are known, gestalt and testing do not always lead to the correct diagnosis. 7,25Fifty-two patients (23%) who met inclusion criteria did not have a LUS performed when one was available on the ambulance.It is possible that a paramedic not trained in LUS was riding in an ambulance that day, but more likely that paramedics trained in LUS chose not to perform LUS.This is similar to prior studies finding issues with paramedic uptake and lower enrollment of patients. 11,12,15It is, thus, possible our results are skewed as the paramedics who felt more confident in using LUS performed the majority of scans in the LUS group.This further highlights that continued LUS training for paramedics is necessary to potentially improve confidence, usage, and appropriate treatment.

CON CLUS IONS
Lung ultrasound improved paramedic sensitivity and accuracy for diagnosing acute heart failure in the prehospital setting when compared to the no lung ultrasound control group.Lung ultrasound use led to higher rates of prehospital heart failure therapy initiation and significantly decreased time to treatment.Future studies are needed with larger sample sizes to verify if these results hold true using randomized methodology.

AUTH O R CO NTR I B UTI O N S
Frances M. Russell and Michael Supples conceived the study, designed the trial, and obtained research funding.All authors supervised the conduct of the trial and data collection.Frances M. Russell drafted the manuscript, and all authors contributed substantially to its revision.Frances M. Russell takes responsibility for the paper as a whole.

CO N FLI C T O F I NTER E S T S TATEM ENT
FMR is a consultant for Butterfly Network.
were followed through hospitalization to assess ED treatment and outcomes.All LUS procedures completed by paramedics during the study period were reviewed by a single expert in LUS with fellowship training in emergency ultrasound, Focused Practice Designation (FPD) in Advanced Emergency Medicine Ultrasonography (AEMUS) credentialing and over a decade of experience researching LUS in patients with AHF.Feedback regarding image quality and interpretation was given via email, on a monthly basis, throughout the study to paramedics submitting images.A random sample of LUS images (20%) were reviewed by a second expert to determine interrater reliability between experts.A 5-point image quality review was used and adopted from prior studies: (1) no recognizable structures; (2) minimally recognizable structures but insufficient for diagnosis; (3) minimal criteria met for diagnosis, recognizable structures with flaws;(4) minimal criteria met for diagnosis, all structures imaged well and diagnosis easily supported; or (5) minimal criteria met for diagnosis, all structures imaged with excellent image quality and diagnosis completely supported.21 Of the 94 patients with AHF, 14% (11/77) received HF therapy prehospital without the use of LUS and 53% (9/17) with the use of LUS.LUS improved significantly improved frequency of treatment (p = 0.002).Median and mean times to treatment were 21 and 73 min (range 12 to 260 min) with LUS and 169 and 152 min (range 12 to 455 min) without LUS.

F I G U R E 1
Patient flow through study.AHF, acute heart failure; FN, false negative; FP, false positive; LUS, lung ultrasound; sn, sensitivity; sp, specificity; TN, true negative; TP, true positive.based on a study by Sforza et al., 20 evaluating accuracy of LUS for AHF based on the location of zones positive for B-lines.Although we did not track time to completion of the LUS study, this protocol was feasible even with shorter transport times as it only included a maximum of four zones and a minimum of two zones, with the median transport time being 13 min.
Patient demographics with and without AHF.Patient demographics with and without LUS.
The other authors declare no conflicts of interest.Abbreviations: ICU, intensive care unit; LUS, lung ultrasound.